Chunking is a basic domain-general cognitive process. It increases the capacity for information processing by fusing co-occurring events into larger units through associative links. Chunking has shaped and continues shaping language structure. We can see the effects of chunking in the vast phraseologies of languages and in the processes of language change, such as grammaticalization. Chunking lies at the heart of many grammatical theories, such as cognitive, usage-based, and construction grammars. In this talk, I will discuss two potential blindspots. First, language is usually described at the level of a community. However, if language is an inventory of patterns learned from exposure, as usage-based approaches suggest, then personal inventories should vary as a function of usage. Second, chunking has mostly been viewed as a mechanism of unit formation. However, to process information in real time, humans also seem to partition incoming stream into temporal groups. This latter process can also allow for substantial individual variation due to characteristic degeneracy and syntagmatic redundancy of language structure. I will present recent results from a series of studies investigating these two chunking mechanisms.